Skip to content
All systems operational0 AI providers monitored, polled every 2 minutes
Live status
All benchmarks

SWE-bench leaderboard

SWE-bench evaluates language models on their ability to resolve real GitHub issues from popular Python repositories. The model is given an issue description and the repository state, and must produce a patch that resolves the issue and passes the project's existing test suite. SWE-bench is the benchmark that most closely tracks "useful for autonomous coding agents" because the tasks are not toy problems, the success criteria is the project's actual tests, and the input footprint forces the model to reason over real-world code at scale.

Current leader
Claude Opus 5(Anthropic)96%

Last refreshed 2026-09-06. 29 models scored on this benchmark.

Full leaderboard

#ModelProviderScoreReleased
1Claude Opus 5Anthropic96%vendor2026-07
2GLM-5.3Z.ai95.4%independent2026-08
3Claude Fable 5Anthropic95%2026-06
4Kimi K3Moonshot AI93.4%independent2026-07
5Claude Opus 4.8Anthropic88.6%2026-05
6Claude Opus 4.7Anthropic87.6%2026-04
7Grok 4.5xAI86.6%independent2026-07
8Qwen3.8-MaxAlibaba85.6%independent2026-08
9Claude Sonnet 5Anthropic85.2%vendor2026-06
10GLM-5.2Z.ai82.8%independent2026-06
11GPT-5.5OpenAI82.6%2026-04
12Claude Opus 4.6Anthropic80.8%2026-03
13DeepSeek V4 ProDeepSeek80.6%2026-04
14MiniMax M3MiniMax80.5%vendor2026-06
15Claude Sonnet 4.6Anthropic79.6%2026-02
16DeepSeek V4 FlashDeepSeek79%2026-04
17InklingThinking Machines77.6%vendor2026-07
18Mistral Medium 3.5Mistral77.6%2026-05
19Muse Glimmer 30BMeta76%vendor2026-08
20Claude Haiku 4.5Anthropic73.3%2026-01
21Gemini 2.5 ProGoogle63.8%2026-01
22Nemotron 3.5 LightningNVIDIA51.56%vendor2026-08
23o3-miniOpenAI49.3%2025-01
24o1OpenAI48.9%2024-12
25Mistral LargeMistral47.2%2025-11
26DeepSeek V3DeepSeek42%2025-12
27GPT-4.5OpenAI38%2025-12
28GPT-4oOpenAI33.2%2024-05
29Llama 4 MaverickMeta24%2025-04

Score interpretation

Scores are reported as resolution rate (% of issues correctly patched). The headline number on TensorFeed is the SWE-bench Verified subset, the human-validated tasks where the test suite has been confirmed to be a fair signal. Mind the gap between sources: the official swebench.com board tops out around 79% and has taken no new submission since early 2026, while vendor launch materials and independent evaluators now report figures in the low to mid nineties on their own harnesses. Both are real measurements of different setups, and the spread between them is wider than the spread between most models. Anything above 60% is a genuinely useful coding agent; above 90% on a vendor harness, check which harness before you compare.

70%+
Frontier-class. Genuinely useful coding agent territory.
50-70%
Production-ready for assisted coding workflows.
30-50%
Useful for narrow tasks but not autonomous agents.
< 30%
Plausible-looking code that often does not work.

Why this matters for AI agents

If you are building a coding agent, this is the benchmark that matters most. Models with high SWE-bench scores produce patches that compile, pass tests, and respect existing patterns in the codebase. Models with low SWE-bench scores produce code that looks plausible but breaks the build.

Other benchmarks

Premium API: time-series for SWE-bench

The leaderboard above is a snapshot. Want to see how a model's SWE-bench score has moved over the last 30-90 days, or set a webhook that fires when a score crosses a threshold? The premium API has both:

SWE-bench source ·Last refreshed 2026-09-06·Max score 100